Multi-Criteria ABC Segmentation

This will demonstrate to you how to perform Multi-Criteria ABC Segmentation using both R-Programming and Python.

Multi-criteria ABC segmentation is a supply chain approach that involves categorizing products depending on a variety of factors such as cost, demand, and revenue. The goal of this method is to increase inventory management efficiency by allowing firms to prioritize resources and allocate them to the most profitable or in-demand products. Businesses may build a more sophisticated and accurate segmentation that takes into consideration a variety of characteristics rather than just one by using numerous criteria.

Businesses must first choose the criteria that are most relevant to their business before implementing multi-criteria ABC segmentation. Product profitability, demand variability, and lead time are examples of such factors. Following the identification of these criteria, products are divided into groups based on their performance across each of these criteria. This division enables organizations to properly allocate resources by emphasizing high-value items and lowering investment in low-value ones.

One of the primary advantages of multi-criteria ABC segmentation is that it allows organizations to make better inventory management decisions. Businesses may limit the risk of stockouts and lost sales by prioritizing high-value products, while simultaneously lowering the expense of carrying inventory. Furthermore, this segmentation can assist organizations in identifying cost-cutting options by highlighting items that are less profitable or have lesser demand. Overall, multi-criteria ABC segmentation is an effective strategy for increasing supply chain efficiency and profit.

In total, there will be two sets of scripts which make use of ABC Analysis to perform segmentation:

• Product Segmentation (via R-Programming)
• Supplier Segmentation (via Python)

Product Segmentation

The ABC analysis is based on the Pareto principle, which argues that the bulk of revenue is generated by a small fraction of the items (or customers) (or profit). Usually, we we will classify these SKUs into either one of the three categories:

Once the SKUs are classified, the strategies for inventory management will be different across them:

• Category A: High-value products that represent a significant proportion of the revenue and profit.
• Category B: Products with moderate value.
• Category C: Low-value products that generate minimal revenue or profit.

This will be all it takes, if you are performing a normal ABC Analysis using one criteria (i.e. revenue). However, if you are looking to use two or more criterias (i.e. both quantity and revenue), then you would have to perform a Multi-Criteria ABC Analysis.

Quantity	Revenue	Category
A	A	Margin & Volume Drivers
A	B	Volume Drivers
A	C	Volume Drivers
B	A	Margin Driver
B	B	Regular
B	C	Regular
C	A	Margin Driver
C	B	Regular
C	C	Slow Moving

Businesses can then develop particular strategies for controlling each category of products after they have been separated. High-value products in the A category, for example, may necessitate more frequent monitoring and replenishment, but low-value products in the C category may benefit from a vendor-managed inventory program. Businesses may enhance their inventory management operations and boost profitability by customizing their tactics to the individual features of each group.

In brief, multi-criteria ABC analysis enables firms to develop a more tailored and specialized set of plans by taking into consideration a variety of aspects that influence inventory performance. By adapting their tactics to the individual demands of each group, firms may optimize their inventory management procedures and drive profitability.

Supplier Segmentation

The Kraljic Matrix is a supply chain management tool that helps firms examine their purchasing portfolio and build supplier sourcing strategy. Peter Kraljic created the matrix in 1983, and it has two dimensions: supply risk and profit effect. The potential of supply interruption is referred to as supply risk, whereas profit impact is the financial impact of the item on the organization. Businesses can establish customized sourcing strategies for each item by putting products on the matrix depending on these two parameters.

The Kraljic Matrix is noteworthy because it gives a structured method to purchasing portfolio analysis and the formulation of sourcing strategies. Businesses can identify which products require more attention and resources and which can be managed more efficiently by categorizing them based on their supply risk and profit impact. This allows firms to create a tailored sourcing strategy for each item, which can assist to decrease costs, improve quality, and reduce supply chain risk.

On top of that, it also provides a platform for firms to consider their supply chain more strategically. Businesses can build more successful sourcing strategies that correspond with their overall company objectives by focusing on the goods that are most vital to the organization. Businesses, for example, may elect to emphasize on long-term relationships with suppliers for high-risk, high-value things, while employing more transactional ties for low-risk, low-value items. This strategic strategy can assist businesses in increasing supply chain efficiency and profitability.

Two metrics are used to evaluate each supplier - the complexity of their products (risk) and their impact on our business (cost). In order to assess the level of risk, a Risk Rank will be utilized, ranging from 0 (least risky) to 2 (most risky):

risk factors	+ 0.0 (N)	+ 0.5 (Y)
availability	regular	seldom
suppliers	many	few
product_complexity	standard	complex
price_stability	stable	unstable

The cost can be determined using the formula provided below:

Cost = (Cost per unit) x (Total units planned for purchase)

Kraljic's Matrix has four quadrants: Strategic Items, Bottleneck Items, Leverage Items, and Non-Critical Items. The following are the tactics used in each quadrant:

Strategic Items: Things that are both vital to the business and have a high level of supply risk are classified as strategic items. Building long-term relationships with suppliers and working cooperatively to mitigate supply chain risk is the method for controlling these issues. This could include collaborating on joint innovation projects, sharing risk and benefit, and engaging in supplier development programs. The goal is to establish a long-term mutually beneficial relationship that provides a steady supply of necessary commodities.

Bottleneck Items: Bottleneck items are those that are vital to the business yet have a minimal supply risk. The management strategy for these items is to optimize the supply chain in order to reduce lead times and assure timely delivery of the commodities. Working with numerous suppliers, boosting safety stock levels, and building contingency plans to manage supply disruptions may be required. The goal is to maintain a consistent supply of vital commodities while reducing supply chain risk.

Leverage Items: Things with a high level of supply risk but are not vital to the business are considered leverage goods. To manage these products, the company's purchasing power is used to negotiate favorable terms with suppliers. Consolidating purchases, creating alternative suppliers, and using competitive bidding to drive down prices are all possibilities. The goal is to lower the cost of non-essential commodities while minimizing the risk of supply disruption.

Non-Critical Items: Products that are neither vital to the business nor have a significant level of supply risk are classified as non-critical. To reduce procurement costs, the strategy for handling these things is to use transactional relationships with suppliers. Using online marketplaces, standardizing requirements, and minimizing the number of vendors may all be part of this. The goal is to reduce the cost of non-critical commodities while eliminating additional supply chain complexity.

Installation

For Part 1 - Download and install these packages from official CRAN repository:

• readr: to read data from .csv or .xlsx files.
• dplyr: to provide a set of functions for data manipulation.
• DT: to allow you to work with large datasets.
• ggplot2: to create a wide variety of plots and charts.

install.packages(c("readr", "dplyr", "DT", "ggplot2"))

For Part 2 - Use the package manager pip to install:

• pandas: to perform data administration by pulling in .csv or .xlsx files.
• numpy: to perform data manipulation by establishing arrays.
• seaborn: to perform data visualisation based on matplotlib.
• inventorize: to calculate for inventory metrics, stock-out, and ABC analysis.

pip install pandas
pip install numpy
pip install statsmodels
pip install matplotlib

Sample Dataset

For Product Segmentation, you can download the sales_data_sample_utf8.csv file from the source folder, which is located here.

On the other hand, for Supplier Segmentation, you can download the supplier_data_sample_utf8.csv file from the source folder, which is located here.

Ensure that the file is in CSV UTF-8 format, to avoid UnicodeDecodeError later on.

CODE EXPLANATION FOR PRODUCT SEGMENTATION (R-PROGRAMMING)

PART 1 - PREPARATION

Lines 6-7:
Ensure that all the necessary libraries are loaded and available for use in this project.

required_libraries <- c("tidyverse", "inventorize", "dplyr", "ggplot2")
lapply(required_libraries, require, character.only = TRUE)

Lines 9-11:
Import and clean the CSV dataframe. Use the cleaned data to create a new CSV dataframe.

sales <- read.csv('https://raw.githubusercontent.com/dwoo-work/multi-criteria-abc-segmentation/main/src/sales_data_sample_utf8.csv')
sales <- unique(sales)
sales_clean <- sales

Lines 13-17:
Create a new CSV dataframe that will be used for both ABC Analysis and Multi-Criteria ABC Analysis.

grouped <- sales_clean %>%
  group_by(PRODUCTCODE) %>%
  summarize(total_sales = sum(QUANTITYORDERED),
            total_revenue = sum(SALES)) %>%
  ungroup()

Lines 19-23:
Create a new CSV dataframe that will be used for Multi-Criteria ABC Analysis on a Store-Level (classification in this case is 'Country').

by_store <- sales_clean %>%
  group_by(COUNTRY, PRODUCTCODE) %>%
  summarize(total_sales = sum(QUANTITYORDERED),
            total_revenue = sum(SALES)) %>%
  ungroup()

PART 2 - ABC ANALYSIS

Lines 29-31:
Create a new dataframe (abc) to perform ABC Analysis. Calculate the cumulative percentage of total sales.

abc <- grouped %>%
  arrange(desc(total_sales)) %>%
  mutate(cum_pct_sales = cumsum(total_sales) / sum(total_sales))

Lines 33-35:
Assign ABC categories based on cumulative percentage of total sales.

abc <- abc %>%
  mutate(Category = ifelse(cum_pct_sales <= 0.8, "A",
                           ifelse(cum_pct_sales <= 0.95, "B", "C")))

Lines 37-41:
Plot the Countplot using ggplot2, which shows the number of A, B, and C category SKUs.

abc_countplot <- ggplot(abc, aes(x = Category)) +
  geom_bar() +
  ggtitle("Number of A, B, and C Category SKUs") +
  xlab("ABC Product Category Type") + ylab("Number of SKUs") +
  theme(plot.title = element_text(hjust = 0.5))

Lines 43-48:
Plot the Barplot using ggplot2, which shows the average quantity ordered for each product code belonging to the A, B, C category types.

abc_barplot <- ggplot(abc, aes(x = Category, y = total_sales)) +
  geom_bar(stat = "summary", fun = "mean") +
  ggtitle("Average Quantity Ordered of A, B, and C Category Product Code") +
  xlab("ABC Product Category Type") +
  ylab("Average Quantity Ordered for each Product Code") +
  theme(plot.title = element_text(hjust = 0.5))

PART 3 - MULTI-CRITERIA ABC ANALYSIS

Lines 54-56:
Create a new dataframe (mc_abc) to perform Multi-Criteria ABC Analysis.

mc_abc <- productmix(grouped$PRODUCTCODE, grouped$total_sales, grouped$total_revenue)
str(mc_abc)
table(mc_abc$product_mix)

Lines 58-62:
Plot the Countplot using ggplot2, which shows the number of SKUs belong to each of the 9 Product Mix.

mc_abc_countplot <- ggplot(mc_abc, aes(x = product_mix)) +
  geom_bar() +
  ggtitle("Number of A_A to C_C Category SKUs") +
  xlab("ABC Product Mix") + ylab("Number of SKUs") +
  theme(plot.title = element_text(hjust = 0.5))

Lines 58-62:
Plot the Barplot using ggplot2, which shows the average quantity ordered for each product code belong to each of the 9 Product Mix.

mc_abc_barplot1 <- ggplot(mc_abc, aes(x = product_mix, y = sales)) +
  geom_bar(stat = "summary", fun = "mean") +
  ggtitle("Average Quantity Ordered of A_A to C_C Category Product Code") +
  xlab("ABC Product Mix") +
  ylab("Average Quantity Ordered for each Product Code") +
  theme(plot.title = element_text(hjust = 0.5))

Lines 71-75:
Plot the Barplot using ggplot2, which shows the average revenue for each product code belong to each of the 9 Product Mix.

mc_abc_barplot2 <- ggplot(mc_abc, aes(x = product_mix, y = revenue)) +
  geom_bar(stat = "summary", fun = "mean") +
  ggtitle("Total Revenue of A_A to C_C Category Product Code") +
  xlab("ABC Product Mix") + ylab("Average Revenue for each Product Code") +
  theme(plot.title = element_text(hjust = 0.5))

PART 4 - MULTI-CRITERIA ABC ANALYSIS ON A STORE-LEVEL

Lines 81-86:
Create a new dataframe (mix_country) to perform Multi-Criteria ABC Analysis on a store-level.

mix_country <- inventorize::productmix_storelevel(by_store$PRODUCTCODE, by_store$total_sales, by_store$total_revenue, by_store$COUNTRY)
product_mix <- mix_country %>%
  dplyr::group_by(storeofsku, product_mix) %>%
  dplyr::summarise(SKU_count = n()) %>%
  dplyr::ungroup() %>%
  dplyr::select(storeofsku, product_mix, SKU_count)

Lines 88-89:
Create a variable for Australia's product mix for analysis.

australia <- subset(product_mix, storeofsku == "Australia")
str(australia)

Lines 91-95:
Plot the Countplot using ggplot2, which shows the number of SKUs belong to each of the 9 Product Mix (for Australia).

aus_countplot <- ggplot(australia, aes(x = product_mix, y = SKU_count)) +
  geom_bar(stat = "identity") +
  ggtitle("Number of A_A to C_C Category SKUs for Australia") +
  xlab("ABC Product Mix") + ylab("Number of SKUs") +
  theme(plot.title = element_text(hjust = 0.5))

CODE EXPLANATION FOR SUPPLIER SEGMENTATION (PYTHON)

PART 1 - PREPARATION

Lines 5-6:
Import the required libraries.

import pandas as pd
import seaborn as sns

Lines 8-12:
Import and clean the CSV dataframe.

supplier = pd.read_csv('https://raw.githubusercontent.com/dwoo-work/multi-criteria-abc-segmentation/main/src/supplier_data_sample_utf8.csv')
supplier = supplier.drop_duplicates()
supplier = supplier.dropna()
supplier.info()
supplier.head()

Lines 14-20:
Convert availability section from N or Y, to its 0.0 (risk factor absent) or 0.1 (risk factor present). Then, add all responsed values from Q1 to Q5, to a consolidated availability column within the dataframe.

supplier['AVAILABILITY_Q1'] = pd.to_numeric(supplier['AVAILABILITY_Q1'].replace(['N', 'Y'], ['0.0', '0.1']))
supplier['AVAILABILITY_Q2'] = pd.to_numeric(supplier['AVAILABILITY_Q2'].replace(['N', 'Y'], ['0.0', '0.1']))
supplier['AVAILABILITY_Q3'] = pd.to_numeric(supplier['AVAILABILITY_Q3'].replace(['N', 'Y'], ['0.0', '0.1']))
supplier['AVAILABILITY_Q4'] = pd.to_numeric(supplier['AVAILABILITY_Q4'].replace(['N', 'Y'], ['0.0', '0.1']))
supplier['AVAILABILITY_Q5'] = pd.to_numeric(supplier['AVAILABILITY_Q5'].replace(['N', 'Y'], ['0.0', '0.1']))

supplier['AVAILABILITY'] = supplier['AVAILABILITY_Q1'] + supplier['AVAILABILITY_Q2'] + supplier['AVAILABILITY_Q3'] + supplier['AVAILABILITY_Q4'] + supplier['AVAILABILITY_Q5']

Lines 22-28:
Convert suppliers section from N or Y, to its 0.0 (risk factor absent) or 0.1 (risk factor present). Then, add all responsed values from Q1 to Q5, to a consolidated suppliers column within the dataframe.

supplier['SUPPLIERS_Q1'] = pd.to_numeric(supplier['SUPPLIERS_Q1'].replace(['N', 'Y'], ['0.0', '0.1']))
supplier['SUPPLIERS_Q2'] = pd.to_numeric(supplier['SUPPLIERS_Q2'].replace(['N', 'Y'], ['0.0', '0.1']))
supplier['SUPPLIERS_Q3'] = pd.to_numeric(supplier['SUPPLIERS_Q3'].replace(['N', 'Y'], ['0.0', '0.1']))
supplier['SUPPLIERS_Q4'] = pd.to_numeric(supplier['SUPPLIERS_Q4'].replace(['N', 'Y'], ['0.0', '0.1']))
supplier['SUPPLIERS_Q5'] = pd.to_numeric(supplier['SUPPLIERS_Q5'].replace(['N', 'Y'], ['0.0', '0.1']))

supplier['SUPPLIERS'] = supplier['SUPPLIERS_Q1'] + supplier['SUPPLIERS_Q2'] + supplier['SUPPLIERS_Q3'] + supplier['SUPPLIERS_Q4'] + supplier['SUPPLIERS_Q5']

Lines 30-36:
Convert product complexity section from N or Y, to its 0.0 (risk factor absent) or 0.1 (risk factor present). Then, add all responsed values from Q1 to Q5, to a consolidated product complexity column within the dataframe.

supplier['PRODUCT_COMPLEXITY_Q1'] = pd.to_numeric(supplier['PRODUCT_COMPLEXITY_Q1'].replace(['N', 'Y'], ['0.0', '0.1']))
supplier['PRODUCT_COMPLEXITY_Q2'] = pd.to_numeric(supplier['PRODUCT_COMPLEXITY_Q2'].replace(['N', 'Y'], ['0.0', '0.1']))
supplier['PRODUCT_COMPLEXITY_Q3'] = pd.to_numeric(supplier['PRODUCT_COMPLEXITY_Q3'].replace(['N', 'Y'], ['0.0', '0.1']))
supplier['PRODUCT_COMPLEXITY_Q4'] = pd.to_numeric(supplier['PRODUCT_COMPLEXITY_Q4'].replace(['N', 'Y'], ['0.0', '0.1']))
supplier['PRODUCT_COMPLEXITY_Q5'] = pd.to_numeric(supplier['PRODUCT_COMPLEXITY_Q5'].replace(['N', 'Y'], ['0.0', '0.1']))

supplier['PRODUCT_COMPLEXITY'] = supplier['PRODUCT_COMPLEXITY_Q1'] + supplier['PRODUCT_COMPLEXITY_Q2'] + supplier['PRODUCT_COMPLEXITY_Q3'] + supplier['PRODUCT_COMPLEXITY_Q4'] + supplier['PRODUCT_COMPLEXITY_Q5']

Lines 38-44:
Convert product complexity section from N or Y, to its 0.0 (risk factor absent) or 0.1 (risk factor present). Then, add all responsed values from Q1 to Q5, to a consolidated product complexity column within the dataframe.

supplier['PRICE_STABILITY_Q1'] = pd.to_numeric(supplier['PRICE_STABILITY_Q1'].replace(['N', 'Y'], ['0.0', '0.1']))
supplier['PRICE_STABILITY_Q2'] = pd.to_numeric(supplier['PRICE_STABILITY_Q2'].replace(['N', 'Y'], ['0.0', '0.1']))
supplier['PRICE_STABILITY_Q3'] = pd.to_numeric(supplier['PRICE_STABILITY_Q3'].replace(['N', 'Y'], ['0.0', '0.1']))
supplier['PRICE_STABILITY_Q4'] = pd.to_numeric(supplier['PRICE_STABILITY_Q4'].replace(['N', 'Y'], ['0.0', '0.1']))
supplier['PRICE_STABILITY_Q5'] = pd.to_numeric(supplier['PRICE_STABILITY_Q5'].replace(['N', 'Y'], ['0.0', '0.1']))

supplier['PRICE_STABILITY'] = supplier['PRICE_STABILITY_Q1'] + supplier['PRICE_STABILITY_Q2'] + supplier['PRICE_STABILITY_Q3'] + supplier['PRICE_STABILITY_Q4'] + supplier['PRICE_STABILITY_Q5']

Lines 46:
Create the Risk Index variable within the CSV dataframe.

supplier['risk_index'] = supplier['AVAILABILITY'] + supplier['SUPPLIERS'] + supplier['PRODUCT_COMPLEXITY'] + supplier['PRICE_STABILITY']

Lines 48-49:
Create the Value variable within the CSV dataframe. Low value refers to value below the median level, and high value refers to value above the median level.

supplier['value'] = supplier['PRICEEACH'] * supplier['QUANTITYORDERED']
supplier.value.describe()

Lines 51-52:
Create a new CSV dataframe (cleaned data).

supplier_clean = supplier.copy()
supplier_clean.info()

PART 2 - UTILISE KRALJIC MATRIX

Lines 57-72:
Create a function that defines each supplier in one of the four Kraljic's Matrix quadrants.

def category (x,y):
    if ((x >= supplier_clean.value.median()) & (y >= 1)):
        return 'strategic'
    if ((x >= supplier_clean.value.median()) & (y < 1)):
        return 'leverage'
    if ((x < supplier_clean.value.median()) & (y < 1)):
        return 'non-critical'
    if ((x < supplier_clean.value.median()) & (y >= 1)):
        return 'bottleneck'

for i in range (supplier_clean.shape[0]):
    supplier_clean.loc[i,'category'] = category(supplier_clean.loc[i,'value'],
                                                supplier_clean.loc[i,'risk_index'])

supplier_clean['risk_index'].value_counts()
supplier_clean.category.value_counts()

Lines 74:
Create a scatter plot for all the suppliers, with different colors for each Kraljic's Matrix quadrants.

kraljic_diagram = sns.scatterplot(x = 'risk_index', y = 'value', data = supplier_clean, hue= 'category').set(title = 'Suppliers Classification using the Kraljic Matrix')

Credit

Sales Data Sample (https://www.kaggle.com/datasets/kyanyoga/sample-sales-data)

License

MIT